基于单纯形搜索法和免疫进化微粒群算法,提出1个求解无约束最优化问题的新的混合算法—单纯形搜索法和免疫进化微粒群算法的混合算法.由于它不需要梯度信息,所以具有易实施、收敛速度快和计算准确的优点.为了证明混合算法能够改进免疫...基于单纯形搜索法和免疫进化微粒群算法,提出1个求解无约束最优化问题的新的混合算法—单纯形搜索法和免疫进化微粒群算法的混合算法.由于它不需要梯度信息,所以具有易实施、收敛速度快和计算准确的优点.为了证明混合算法能够改进免疫进化微粒群算法的性能,首先利用6个测试函数进行仿真计算比较,计算结果表明,新的混合算法在求解质量和收敛速率上都优于其它进化算法(IEPSO,PSOPC,GSPSO,LSPSO and CPSO);其次,将新混合算法和最新的3种混合算法进行鲁棒性分析比较,结果表明,新混合算法在解的搜索质量、效率和关于初始点的鲁棒性方面都优于其它算法.展开更多
In order to study the problem that particle swarm optimization (PSO) algorithm can easily trap into local mechanism when analyzing the high dimensional complex optimization problems, the optimization calculation using...In order to study the problem that particle swarm optimization (PSO) algorithm can easily trap into local mechanism when analyzing the high dimensional complex optimization problems, the optimization calculation using the information in the iterative process of more particles was analyzed and the optimal system of particle swarm algorithm was improved. The extended particle swarm optimization algorithm (EPSO) was proposed. The coarse-grained and fine-grained criteria that can control the selection were given to ensure the convergence of the algorithm. The two criteria considered the parameter selection mechanism under the situation of random probability. By adopting MATLAB7.1, the extended particle swarm optimization algorithm was demonstrated in the resource leveling of power project scheduling. EPSO was compared with genetic algorithm (GA) and common PSO, the result indicates that the variance of the objective function of resource leveling is decreased by 7.9%, 18.2%, respectively, certifying the effectiveness and stronger global convergence ability of the EPSO.展开更多
文摘基于单纯形搜索法和免疫进化微粒群算法,提出1个求解无约束最优化问题的新的混合算法—单纯形搜索法和免疫进化微粒群算法的混合算法.由于它不需要梯度信息,所以具有易实施、收敛速度快和计算准确的优点.为了证明混合算法能够改进免疫进化微粒群算法的性能,首先利用6个测试函数进行仿真计算比较,计算结果表明,新的混合算法在求解质量和收敛速率上都优于其它进化算法(IEPSO,PSOPC,GSPSO,LSPSO and CPSO);其次,将新混合算法和最新的3种混合算法进行鲁棒性分析比较,结果表明,新混合算法在解的搜索质量、效率和关于初始点的鲁棒性方面都优于其它算法.
基金Project(70671040) supported by the National Natural Science Foundation of China
文摘In order to study the problem that particle swarm optimization (PSO) algorithm can easily trap into local mechanism when analyzing the high dimensional complex optimization problems, the optimization calculation using the information in the iterative process of more particles was analyzed and the optimal system of particle swarm algorithm was improved. The extended particle swarm optimization algorithm (EPSO) was proposed. The coarse-grained and fine-grained criteria that can control the selection were given to ensure the convergence of the algorithm. The two criteria considered the parameter selection mechanism under the situation of random probability. By adopting MATLAB7.1, the extended particle swarm optimization algorithm was demonstrated in the resource leveling of power project scheduling. EPSO was compared with genetic algorithm (GA) and common PSO, the result indicates that the variance of the objective function of resource leveling is decreased by 7.9%, 18.2%, respectively, certifying the effectiveness and stronger global convergence ability of the EPSO.
